All
AI Agent Governance Glossary
The terms security, platform, and IT teams actually need to govern AI agents in production.
AI Agent Governance
AI Agent Governance
AI agent governance is the practice of controlling what autonomous, tool-using AI agents are allowed to do across their lifecycle: giving each agent an identity, scoping its permissions, enforcing policy at runtime, and logging every action. It is distinct from model safety, which governs what a model says rather than what an agent does in production.
Shadow AI
Shadow AI
Shadow AI is any AI tool, assistant, agent, or MCP connection an employee runs without security review or approval. It spreads because AI adoption outpaces policy, leaving usage that security teams cannot see, scope, or audit.
Model Context Protocol (MCP)
Model Context Protocol (MCP)
The Model Context Protocol (MCP) is an open standard that lets an AI agent discover and call external tools in a uniform way, similar to an API. It moves data between an agent and a tool but enforces no authentication or authorization by default, which is why enterprises add a governance layer on top.
AI Security
AI Security
AI security focuses on protecting artificial intelligence systems, models, and the data they rely on from manipulation, misuse, and evolving cyber threats. This section explores key approaches to securing AI and using AI securely within enterprise environments, from safeguarding training data to mitigating prompt injection and model exploitation risks.
MCP Gateway
MCP Gateway
An MCP gateway is a single control point that sits between every AI agent and every tool it connects to. It authenticates each connection, decides what the agent is allowed to do, and records every call, collapsing scattered agent-to-tool connections into one governed entry point.
Network Security
Network Security
Network security focuses on protecting the integrity, confidentiality, and accessibility of data as it travels across networks. This section provides insights into key principles, technologies, and best practices to help build a secure and resilient network.
Cybersecurity
Cybersecurity
Cybersecurity focuses on safeguarding digital systems, networks, and sensitive data from unauthorized access, disruption, and evolving threats. This section offers insights into key cybersecurity principles.
Cloud Security
Cloud Security
Cloud security involves protecting data, applications, and services hosted in cloud environments from unauthorized access, breaches, and other cyber threats. This section explores key strategies to ensure a safe and reliable cloud infrastructure.
Excessive Agency (OWASP LLM06)
Excessive Agency (OWASP LLM06)
Excessive Agency is an OWASP LLM Top 10 risk (LLM06) covering the damage an AI agent can do when it has more functionality, permissions, or autonomy than its task requires. It is an access problem, not a model problem: the fix is scoped tools, least-privilege permissions, and a human in the loop for high-impact actions.
Runtime Guardrails
Runtime Guardrails
Runtime guardrails evaluate and enforce policy on an AI agent's actions in real time, at the moment of execution, rather than reviewing logs after the fact. They can allow low-risk actions automatically, block disallowed ones, and pause high-risk actions for human approval before anything runs.
Prompt Injection
Prompt Injection
Prompt injection is an attack that hides malicious instructions in content an AI agent processes, such as a document, email, or tool response, so the agent executes them as legitimate tasks using its own access. Indirect prompt injection needs no direct system access, only content the agent will read.
Agent Identity (Non-Human Identity)
Agent Identity (Non-Human Identity)
Agent identity gives an AI agent its own accountable identity, distinct from the human who launched it but tied back to that person through the identity provider. These non-human identities can outnumber human ones many times over, and without them there is no way to attribute an agent's actions or revoke its access cleanly.
AI Security Policies: Structure and Template
Learn how to structure AI security policies, what controls to include, and how to build a practical policy template for governing enterprise AI.
AI Security Posture Management (AI-SPM): Definition, Risks, and Best Practices
Learn what AI Security Posture Management (AI-SPM) is, the risks it helps uncover, and best practices for improving visibility, governance, and control across enterprise AI.
SOC 2 for AI: The Complete Compliance Guide
Learn how SOC 2 applies to AI systems, which controls matter most, and how enterprises can prepare AI operations for security and compliance requirements.
SOC 1 for AI: Compliance Requirements and Controls
Learn what SOC 1 means for AI systems, which controls matter, and how enterprises can prepare AI operations for compliance and audit requirements
Shadow AI Detection: Find Unmanaged AI in Your Enterprise
Learn how shadow AI detection helps enterprises discover unmanaged AI tools, identify hidden risks, and bring AI usage under control.
What Is an AI Audit Trail? Logging, Monitoring, & Compliance
Learn how enterprises manage AI tool catalogs, control tool access, and enforce governance over the tools AI agents can discover and use.
AI Tool Management: How Enterprises Control AI Tool Catalogs
Learn how enterprises manage AI tool catalogs, control tool access, and enforce governance over the tools AI agents can discover and use.
Tool Governance: Definition, Challenges, Strategies for AI
Learn what tool governance means for AI, the key challenges it creates, and strategies for controlling how AI agents access and use enterprise tools.
Model Context Protocol (MCP): Fundamentals Explained
Learn what the Model Context Protocol (MCP) is, how it works, and how it helps AI agents securely connect to tools, data, and external systems.
What Is an MCP Gateway? How It Works and Why It Matters
Learn what an MCP gateway is, how it connects AI agents to tools and data, and why it matters for security, control, and enterprise governance.
AI Risk Management: Frameworks, Process & Best Practices
Learn how AI risk management works, the frameworks and processes enterprises use, and best practices for identifying, assessing, and reducing AI risk.
AI Usage Control: AI Usage Control, Definition and How It Works
AI usage control finds, scores, and governs how employees and AI agents use AI. See how it works, how it differs from DLP, and what to look for in a tool.
AI Governance Framework: Structure for Trust and Controls
Learn how an AI governance framework helps enterprises build trust, manage risk, and establish clear controls for secure AI adoption.
Zero Trust for AI Agents: Principles, Risks, and Best Practices
Learn how Zero Trust applies to AI agents, the key risks it addresses, and best practices for securing agent access, permissions, and actions.
Agent Roles: How Role-Based Access Works for AI Agents
Learn how role-based access works for AI agents, how agent roles define permissions, and how enterprises can enforce safer, more controlled access.
Agent Permissions: Definition, Levels, and How They Work
Learn what AI agent permissions are, the different permission levels, and how they control what agents can access and do.
Agent Authorization: Tool-Level vs. Action-Level Permission
Learn the difference between tool-level and action-level authorization for AI agents, and how granular permissions improve security and control.
Least Privilege for AI Agents: Security Best Practices
Learn how least privilege helps secure AI agents by limiting access, permissions, and actions to only what they need.
AI Governance Framework: Structure for Trust and Controls
A practical AI governance framework for building trust, control, accountability, and secure AI adoption across the enterprise.
What Is an MCP Server? How It Works and What It Does
Learn what an MCP server is, how it connects AI agents to tools and data, and why it matters for secure, scalable AI infrastructure.
AI Access Control: Definition, How It Works, and Why Enterprises Need It
How AI access control gives every agent the right permissions, limits risky actions, and helps enterprises govern AI at scale.
Agent Authentication: Definition, How It Works, and Security Risks
Learn what agent authentication is, how AI agents prove their identity, and the security risks of weak or mismanaged authentication.
What Is Agent Identity? How Identity Works for AI Agents
Learn what agentic AI is, how it works, where it’s used, and the key security risks organizations need to manage.
Agentic AI: Definition, Use Cases, and Risks
Learn what agentic AI is, how it works, its key use cases and benefits, and the security risks that come with autonomous AI agents.
What is a CISO? Roles and Responsibilities
Learn what a CISO does, their key roles and responsibilities, how they differ from a CIO, and how AI agents are reshaping security, governance, and risk management.
What Is an Exposure Window in Cybersecurity?
An exposure window is the period of time when a specific weakness is present, reachable, and not yet effectively mitigated. In plain terms, it is how long an organization remains open to attack because of a vulnerability, misconfiguration, leaked credential, overprivileged identity, exposed service, or other exploitable condition.
The term is useful because it adds time to the risk discussion. A vulnerability describes the weakness. An exposure describes the weakness in a reachable or insufficiently protected state. The exposure window describes how long that state lasts.
AI Security for Applications: Definition, Threats, and Controls
AI security for applications is the practice of protecting software that uses AI from attacks, misuse, data exposure, and unsafe behavior. It covers the application code, model behavior, prompts, training and retrieval data, inference APIs, connected tools, and any AI agents that can take action.
What is AIOps (Artificial Intelligence for IT Operations)?
AIOps is the use of artificial intelligence and machine learning to improve IT operations. In practical terms, it helps operations, platform, DevOps, and SRE teams make sense of large volumes of logs, metrics, traces, events, tickets, and infrastructure data so they can spot issues earlier, reduce alert noise, find likely causes faster, and automate routine response work.
What Is Virtual Patching and How Does It Reduce Cybersecurity Risk?
Virtual patching is an interim security control that blocks or filters attempts to exploit a known weakness without changing the vulnerable software itself. Instead of editing source code, installing a vendor update, or modifying the affected package, defenders place an enforcement layer in front of the asset and use rules, signatures, or policy logic to stop the exploit path.
Your agents are already in the wild.
Give them a Basecamp. Go from AI chaos to AI work, in minutes.